基于改进的协方差矩阵描述子的点云配准方法

doi:10.16182/j.issn1004731x.joss.22-0010

系统仿真学报 ›› 2023, Vol. 35 ›› Issue (5): 979-986.doi: 10.16182/j.issn1004731x.joss.22-0010

基于改进的协方差矩阵描述子的点云配准方法

张元¹^,²^,³(), 韩浩宇¹^,²^,³(), 韩燮¹^,²^,³, 付嘉旭¹^,²^,³

^1.中北大学计算机科学与技术学院，山西太原 030051
^2.机器视觉与虚拟现实山西省重点实验室，山西太原 030051
^3.山西省视觉信息处理及智能机器人工程研究中心，山西太原 030051

收稿日期:2022-01-06 修回日期:2022-02-28 出版日期:2023-05-30 发布日期:2023-05-22
通讯作者: 韩浩宇 E-mail:zhangyuan@nuc.edu.cn;985811696@qq.com
作者简介:张元(1976-)，女，副教授，硕士，研究方向为计算机视觉。E-mail：zhangyuan@nuc.edu.cn
基金资助:
国家重点研发计划(2018YFB2101504);山西省重点研发计划(201903D121147);山西省自然科学基金(201901D111150)

Point Cloud Registration Method Based on Improved Covariance Matrix Descriptor

Yuan Zhang¹^,²^,³(), Haoyu Han¹^,²^,³(), Xie Han¹^,²^,³, Jiaxu Fu¹^,²^,³

^1.Computer Science and Technology Department, North University of China, Taiyuan 030051, China
^2.Shanxi Provincial Key Laboratory of Machine Vision and Virtual Reality, Taiyuan 030051, China
^3.Shanxi Province Visual Information Processing and Intelligent Robot Engineering Research Center, Taiyuan 030051, China

Received:2022-01-06 Revised:2022-02-28 Online:2023-05-30 Published:2023-05-22
Contact: Haoyu Han E-mail:zhangyuan@nuc.edu.cn;985811696@qq.com

摘要/Abstract

摘要：

点云配准是数字化文物保护的关键一环，提高配准精度与抗噪性是文物点云配准的主要目标。针对该问题，提出一种基于协方差矩阵描述子的三维点云配准方法。使用张量投票法剔除噪声点，对剔除噪声后的点云，使用内在形状签名法提取关键点；对提取的关键点构建邻域信息，利用该信息建立协方差矩阵描述子；通过计算最近距离寻找匹配点对，使用法向量夹角对其约束，剔除误匹配点对；选取匹配点对计算变换矩阵完成粗配准，再通过迭代最近点对方法进行精配准。实验结果表明，相比于常用配准算法，本文算法配准精度更高，且适用于低重叠率模型与含噪声模型。

关键词: 点云配准, 协方差矩阵描述子, 关键点, 张量投票, 内在形状签名

Abstract:

Point cloud registration is a key part of the digital protection of cultural relics. Improving registration accuracy and noise resistance is the main goal of point cloud registration for cultural relics. In order to solve this problem, a three-dimensional (3D) point cloud registration method based on a covariance matrix descriptor is proposed. The tensor voting method is used to eliminate the noise points, and the internal shape signature method is used to extract the key points from the point cloud after removing the noise. Then, the neighborhood information is constructed for the extracted key points, and the covariance matrix descriptor is established by using the information. In addition, the matching point pair is found by calculating the nearest distance, and the angle constraint of the normal vector is used to eliminate the wrong matching point pair. The matching point pair is selected, and the transformation matrix is calculated to complete the rough registration. Then the iterative nearest point method is used for the fine registration. Experimental results show that compared with common registration algorithms, the algorithm proposed in this paper has higher registration accuracy and is suitable for models with low overlap rates and noisy models.

Key words: point cloud registration, covariance matrix descriptor, key points, tensor voting, intrinsic shape signature

中图分类号:

TP391.9

张元, 韩浩宇, 韩燮, 付嘉旭. 基于改进的协方差矩阵描述子的点云配准方法[J]. 系统仿真学报, 2023, 35(5): 979-986.

Yuan Zhang, Haoyu Han, Xie Han, Jiaxu Fu. Point Cloud Registration Method Based on Improved Covariance Matrix Descriptor[J]. Journal of System Simulation, 2023, 35(5): 979-986.

图/表 9

图1

图2

图3

图4

图5

图6

图7

表1

表2

参考文献 20

1	熊风光, 蔡晋茹, 况立群, 等. 三维点云模型中特征点描述子及其匹配算法研究[J]. 小型微型计算机系统, 2017, 38(3): 640-644.
	Xiong Fengguang, Cai Jinru, Kuang Liqun, et al. Study on Descriptor and Matching Algorithm of Feature Point in 3D Point Cloud[J]. Journal of Chinese Computer Systems, 2017, 38(3): 640-644.
2	李霞, 熊风光, 李定主, 等. 基于协方差矩阵的多尺度特征描述子[J]. 计算机工程与设计, 2019, 40(3): 795-800.
	Li Xia, Xiong Fengguang, Li Dingzhu, et al. Multi-scale Feature Descriptor Based on Covariance Matrix[J]. Computer Engineering and Design, 2019, 40(3): 795-800.
3	赵明富, 黄铮, 宋涛, 等. 融合采样一致性和迭代最近点算法的点云配准方法[J]. 激光杂志, 2019, 40(10): 45-50.
	Zhao Mingfu, Huang Zheng, Song Tao, et al. Point Cloud Registration Method Based on Sample Consensus Initial Alignment and Iterative Closest Point Algorithm[J]. Laser Journal, 2019, 40(10): 45-50.
4	孙文潇, 王健, 梁周雁, 等. 法线特征约束的激光点云精确配准[J]. 武汉大学学报(信息科学版), 2020, 45(7): 988-995.
	Sun Wenxiao, Wang Jian, Liang Zhouyan, et al. Accurate Registration of Laser Point Cloud with Normal Feature Constraints[J]. Geomatics and Information Science of Wuhan University, 2020, 45(7): 988-995.
5	杨稳, 周明全, 郭宝, 等. 基于曲率图的颅骨点云配准方法[J]. 光学学报, 2020, 40(16): 1-11.
	Yang Wen, Zhou Mingquan, Guo Bao, et al. Skull Point Cloud Registration Based on Curvature Maps[J]. Acta Optica Sinica, 2020, 40(16): 1-11.
6	张元, 李晓燕, 韩燮. 一种低重叠率的三维点云配准方法[J]. 激光与光电子学进展, 2021, 58(8): 170-179.
	Zhang Yuan, Li Xiaoyan, Han Xie. A 3D Point Cloud Registration Method with Low Overlap Rate[J]. Laser & Optoelectronics Progress, 2021, 58(8): 170-179.
7	Serafin Jacopo, Grisetti Giorgio. Using Extended Measurements and Scene Merging for Efficient and Robust Point Cloud Registration[J]. Robotics and Autonomous Systems (S0921-8890), 2017, 92: 91-106.
8	Mohammed Lamine Tazir, Gokhool Tawsif, Checchin Paul, et al. CICP: Cluster Iterative Closest Point for Sparse-Dense Point Cloud Registration[J]. Robotics and Autonomous Systems (S0921-8890), 2018, 108: 66-86.
9	Chang Seunggyu, Ahn Chanho, Lee Minsik, et al. Graph-matching-Based Correspondence Search for Nonrigid Point Cloud Registration[J]. Computer Vision and Image Understanding (S1077-3142), 2020, 192: 102899.
10	Sánchez-Belenguer Carlos, Ceriani Simone, Taddei Pierluigi, et al. Global Matching of Point Clouds for Scan Registration and Loopdetection[J]. Robotics and Autonomous Systems (S0921-8890), 2020, 123: 109963.
11	Philipp Bauer, Lars Heckler, Mario Worack, et al. Registration Strategy of Point Clouds Based on Region-Specific Projections and Virtual Structures for Robot-Based Inspection Systems[J]. Measurement (S0263-2241), 2021, 185: 103324.
12	Mirza Tahir Ahmed, Ziauddin Sheikh, Marshall Joshua A., et al. Point Cloud Registration Using Virtual Interest Points from Macaulay's Resultant of Quadric Surfaces[J]. Journal of Mathematical Imaging and Vision (S1573-7683), 2021, 63: 457-471.
13	魏亮, 薛牧遥, 霍炬, 等. 基于 ICP 算法的非合作目标特征点云配准优化[J]. 系统仿真学报, 2020, 32(12): 2383-2387.
	Wei Liang, Xue Muyao, Huo Ju, et al. Non-Cooperative Target Feature Point Cloud Registration Optimization Based on ICP Algorithm[J]. Journal of System Simulation, 2020, 32(12): 2383-2387.
14	张旭春, 周洪军, 郑津津, 等. 基于多尺度特征及点距离约束的点云配准[J]. 激光与光电子学进展, 2021, 58(24): 2415004.
	Zhang Xuchun, Zhou Hongjun, Zheng Jinjin, et al. Point Cloud Registration Based on Multi-scale Feature and Point Distance Constraint[J]. Laser & Optoelectronics Progress, 2021, 58(24): 2415004.
15	孙水发, 李准, 夏坤, 等. 变尺度点云配准算法[J].系统仿真学报, 2018, 30(7): 2466-2474.
	Sun Shuifa, Li Zhun, Xia Kun, et al. Variable Scale Point Cloud Registration Algorithm[J]. Journal of System Simulation, 2018, 30(7): 2466-2474.
16	Yu Zhong. Intrinsic Shape Signatures: A Shape Descriptor for 3D Object Recognition[C]// IEEE 12th International Conference on Computer Vision Workshops. Kyoto, Japan: IEEE, 2009: 689-696.
17	韩浩宇, 张元, 韩燮. 一种改进的激光点云滤波算法[J]. 激光与光电子学进展, 2021, 58(20): 93-99.
	Han Haoyu, Zhang Yuan, Han Xie. An Improved Filtering Method of Laser Point Cloud[J]. Laser & Optoelectronics Progress, 2021, 58(20): 93-99.
18	Liu Zheng, Xiao Xiaowen, Zhong Saishang, et al. A Feature-Preserving Framework for Point Cloud Denoising[J]. Computer-Aided Design (S0010-4485), 2020, 127: 102857.
19	熊风光. 三维点云配准技术研究[D]. 太原: 中北大学, 2018.
	Xiong Fengguang. Research on Registration Technology of 3D Point Cloud[D]. Taiyuan: North University of China, 2018.
20	Zampogiannis Konstantinos, Fermuller Cornelia, Aloimonos Yiannis. Topology-Aware Non-Rigid Point Cloud Registration[J]. IEEE Transaction on Pattern Analysis and Machine Intelligence (S0162-8828), 2021, 43(3): 1056-1069.

点云模型	Go-ICP	RANSAC+ICP	文献[1]方法	本文方法
斯坦福bun045 and bun090	0.010 6	0.009 8	0.007 1	0.006 4
斯坦福bun000 and bun090	0.113 0	0.057 5	0.021 3	0.010 7
斯坦福dragon000 and dragon048	0.005 4	0.005 3	0.005 1	0.004 4
青铜鸟尊	1.05	0.93	0.71	0.55
唐三彩模型	1.97	1.51	1.46	1.38

点云模型	Go-ICP	RANSAC+ICP	文献[1]方法	本文方法
斯坦福bun045 and bun090	73.24	69.81	128.63	57.85
斯坦福bun000 and bun090	93.31	63.27	127.19	57.99
斯坦福dragon000 and dragon048	81.57	55.63	125.74	51.54
青铜鸟尊	72.05	63.35	126.48	53.31
唐三彩模型	113.15	89.23	163.95	77.06

[1]	余文利, 周明全, 税午阳, 武仲科. 基于曲率的点云自动配准方法[J]. 系统仿真学报, 2015, 27(10): 2374-2379.
[2]	陈明伟, 杨洋. 基于张量投票和深度图的弱纹理图像三维重构[J]. 系统仿真学报, 2015, 27(10): 2553-2559.

基于改进的协方差矩阵描述子的点云配准方法

Point Cloud Registration Method Based on Improved Covariance Matrix Descriptor

RichHTML

PDF (PC)

可视化

摘要/Abstract

引用本文

使用本文

图/表 9

参考文献 20

相关文章 2

编辑推荐

Metrics

本文评价